The invention relates to a circuit for providing a high-value resistance in an integrated circuit and, in particular, to a circuit for multiplying the resistance value of a resistor so that a high-value resistance can be provided in less area and with stable operation characteristics.
In integrated circuit designs, a high-value resistance is often needed. Directly creating such a high-value resistance element may be either unfeasible or undesirable due to tradeoffs that need to be made in performance and/or size of such a resistance element. For example, one conventional method for realizing high-value resistances in an integrated circuit includes using a silicon resistance such as the base region of an NPN bipolar transistor to form the resistance element. While using the NPN base region offers the best accuracy and best thermal and voltage coefficients, typical resistivities are so low as to make high-value resistances very area-consuming. For instance, a typical sheet resistance of the base region is 2 kohm/square. To obtain a 500 kohm resistor will require 250 squares. Not only is the area consumed by 250 squares significant, but the capacitance associated with the resulting resistance element is undesirably large, causing significant degradation in the element's frequency response.
Another convention method for realizing high-value resistances in an integrated circuit involves using a weakly-enhanced, long-channel length MOS transistor. Although such as resistance element may provide a more compact realization than the previous method, the resistance value will vary more widely over process condition and temperature variations.
Finally, another conventional method for realizing high-value resistance involves using a high-resistivity material such as a lightly-doped polysilicon layer or the silicon well region. The disadvantage with using either of these structures is the large temperature coefficient and large voltage coefficient of resistance exhibited by resistors made of these materials.
The conventional methods for providing high-value resistance in an integrated circuit are undesirable as they either require large silicon area to implement or the resulting resistance element exhibits large resistance variations due to processing conditions, operating temperature, and voltage variations. Therefore, a method for providing a high-value resistance in an integrated circuit without the aforementioned shortcomings is desired.
According to one embodiment of the present invention, a circuit coupled to a first node of a first circuit includes a first transistor, a second transistor being N times larger than the first transistor, and a resistor for providing a resistance value at the first node that is a multiple of the resistance value of the resistor. The first transistor has a control terminal and a first current handling terminal coupled to the first node and a second current handling terminal coupled to a second node. The second transistor has a control terminal coupled to the first node, a first current handling terminal coupled to a first supply voltage, and a second current handling terminal coupled to the second node. The resistor is coupled between the second node and a second supply voltage.
As thus configured, the first and second transistors form a current mirror with the second transistor draws N times more current than the first transistor. In operation, when a voltage is applied to the first node, a resistance value being (N+1) times the resistance of the resistor is established at the first node.
The first and second transistors can be implemented using NPN bipolar transistors, PNP bipolar transistors, NMOS transistors, or PMOS transistors.
The present invention is better understood upon consideration of the detailed description below and the accompanying drawings.
In accordance with the principles of the present invention, a circuit for providing a high-value resistance to a first node uses active circuitry to multiply the resistance value of a resistor. In this manner, a high-value resistance can be provided in an integrated circuit by using a comparatively small structure. In one embodiment, the resistance multiplier circuit includes a pair of unevenly sized transistors coupled to the first node and to a resistor. The transistors are configured as a current mirror for drawing currents through the resistor. By discarding part of the current drawn, the resistance value as seen from the first node can be made larger than actual resistance of the resistor itself. The geometric ratio of the pair of transistors establishes the amount of resistance multiplication that can be realized.
The resistance multiplier circuit of the present invention offers many advantages. First, the resistance multipler circuit including two transistors requires less silicon area to implement while providing for very high resistance values. Second, the circuit can achieve stability over process variations and over operating temperature and voltage variations. Lastly, because the circuit can achieve large resistance in a small area, the parasitic capacitance of the resulting structure is much lower than that of the conventional structures described above.
A resistor R is coupled to the emitter terminals of transistors Q1 and Q2 (node 14) and a virtual ground node (node 18). Virtual ground node 18 can be the ground potential or a negative supply voltage of circuit 12. The emitter currents from transistors Q1 and Q2 flow into resistor R and a voltage VR is established at node 14. The current flowing through resistor R is given by IR=VR/R. Resistance multiplier circuit 10 operates to provide a multiple of the resistance of resistor R at node X where the amount of multiplication is determined by the geometric ratio of transistors Q1 and Q2. Specifically, the resistance as seen from node X is (N+1) times the resistance of R, as will be explained below. Because the area ratio N is a fixed number and does not vary with operating temperature, operating voltages or processing conditions, the high-value resistance obtained at node X as a result of multiplying resistor R exhibits stable operating characteristics over temperature, voltage and process condition variations.
In operation, it is assumed that node X presents a voltage Vx to the base terminals of transistors Q1 and Q2. As a result of the application of voltage Vx, transistors Q1 and Q2 are turned on and emitter currents from each transistor flow into resistor R. Thus, current IR represents the sum of the emitter currents from transistors Q1 and Q2. Specifically, if an emitter current IE flows in transistor Q1, an emitter current N*IE will flow in transistor Q2. Consequently, resistor current IR is (N+1)IE.
Resistance multiplier circuit 10 splits the resistor current IR through resistor R and “throws away” a fraction of the resistor current to a convenient system potential, such as the virtual ground node 18. The portion of current (current Ix) not diverted to the virtual ground node is directed into node X to which a high resistance is established. In accordance with the present invention, by virtue of connecting the collector terminal of transistor Q2 to the supply voltage, the current in transistor Q2 is being discarded while the current in transistor Q1 is being directed into node X as current Ix. Thus, the fraction of current IR being thrown away is the current in transistor Q2 and is given by N/(N+1), where N is the emitter area ratio of transistors Q1 and Q2.
Node X, therefore, sources a small signal current given by:
Ix=Vx/(R*(N+1)).
According to the above relationship, the effective small-signal resistance looking out from node X is approximately:
Rx=R*(N+1).
The above equation holds if the small-signal emitter impedance of transistor Q1 is much smaller than the resistance R. This condition is met if the voltage drop across the resistor R (voltage VR) is much greater than 1/N times the thermal voltage VT given by VT=kT/q (26 mV at 300K).
As a result of developing a current of (N+1)IE at resistor R and throwing away N/(N+1) amount of the resistor current in transistor Q2, a high-value resistance is established at node X where the resistance Rx is (N+1) times the resistance of resistor R. In this manner, a very high resistance value can be provided to circuit 12 by using a relatively small resistor R.
The operational principles of resistance multiplier circuit 10 can be examined using the following small signal analysis.
Applying Kirchhoff's Current Law to the emitter node (node 14′) of the small signal circuit, the following equation results:
ve/ro−gm2(vt−ve)+ve/R−it=0. Eq. (1)
Collecting terms and substituting the following terms for gm2 and ve:
gm2=N*gm1 and
ve=vt−(it/gm1),
the following equation for a resistance rt at the input node is derived:
Equation (2) above can be simplified by applying the following assumptions. First, the output impedance ro of transistor Q2 is assumed to be much greater than the discrete resistor R. The output impedance of transistor Q2 has to be large so that the current drawn from the test excitation source vt is virtually equal to the current through the resistor R. By making the resistance of resistor R small in comparison to the output impedance ro, the output impedance ro draws insignificant amounts of current compared to the resistance R.
Second, the combined emitter conductance (N+1)*gm1 of the two transistors is assumed to be much greater than the discrete conductance 1/R. In other words, the small signal emitter resistance of the two transistors Q1 and Q2 should be much smaller than the resistance of resistor R so that the emitters of the transistors can drive the voltage on the emitter node 14′. In this manner, the signal at the emitter terminals of the transistors will track the signal at the base terminals in the AC mode. If the above two conditions are met, then the AC signals impressed at the input node (such as voltage vt) result in a current change that is determined only by resistor R and not by the output impedance of transistor Q2.
Applying the above two assumptions to Equation (2), the equation can be simplified as follows. First by applying the assumption ro>>R and collecting terms:
Then by applying the assumption (N+1)gm1>>1/R or (N+1)R>>1/gm1, the intuitive result for resistance rt at the input node of the resistance multiplier circuit is obtained:
rt=(N+1)R.
Note that the above assumptions can be met as long as a resistance value for resistor R is chosen so that the voltage drop across resistor R is much larger than the thermal voltage VT (or kT/q).
The small signal circuit analysis above illustrates that an effective resistance Rx that is (N+1) times the resistance of the resistor R can be realized at the input node X by using the resistance multiplier circuit of the present invention.
In the operation of resistance multiplier circuit 10, a portion of the supply current generated is thrown away. Specifically, N/(N+1) portion of the current drawn out of the resistor R is discarded. Therefore, the resistance multiplier circuit of the present invention increases the overall supply current for circuit 12 as compared to the case when a direct realization of a high resistance element is used. However, in many practical cases, the extra “throw-away” current will not be large enough to present a problem to the operation of circuit 12 as the current drawn by a high-value resistance is, by definition, very small and therefore the “throw-away” current is not very large in comparison.
In the present embodiment, resistor R is fabricated as a base diffusion resistor in an integrated circuit. In other embodiments, other structures, such as a polysilicon layer or an enhancement mode MOS transistor, can be used to form resistor R, as is well understood by one skilled in the art.
The use of NPN bipolar transistors in resistance multiplier circuit 10 is illustrative only. The resistance multiplier circuit of the present invention can also be constructed using other transistor devices depending on the application.
Of course, in another embodiment, the resistance multiplier circuit of the present invention can be implemented using PMOS transistors in place of the PNP bipolar transistors in
In the circuit arrangement of
In the embodiment shown in
ro<<Reff/(abs(N2−N1))
where N2 and N1 are the size ratios of the NPN transistor pairs, Reff is the ideal multiplied resistance, and ro is the output resistance of the bias current source. This condition ensures that the size ratio mismatch does not result in a significant voltage modulation across the bias source. If a significant voltage modulation were to happen, the current through the resistors R1 and R2 would not be solely determined by the voltage impressed across multiplier circuits 40A and 40B. In any case, the desired size ratios for the pair of resistance multiplier circuits can be selected to obtain the desired resistance value between nodes X1, X2, which resistance values do not have to be equal.
According to one aspect of the present invention, the resistance multiplier circuit is coupled to an amplifier circuit to function as a gain attenuator. Because the resistance multiplier circuit of the present invention can provide a large resistance value in a very small area, effective gain attenuation can be realized in a small area with very stable operating characteristics.
In amplifier circuits, it is quite often advantageous to limit the gain of the amplifier to make compensation easier. Gain attenuation can be done by degenerating the input structure. But common methods of degenerating the gain structure introduce offset voltages that can be significant. Thus, in some cases, it would be preferable to decrease the output impedance of the amplifier, such as by attaching a resistor having a resistance value somewhat lower than the nodal dynamic resistance of the output of the amplifier. But the appropriate value of such a resistor is typically in the hundreds of kilo-Ohms. When the conventional methods are used to implement such a resistor having a large resistance, the resulting resistor structure is large in size and has associated with it a high parasitic capacitance. Large resistor size is undesirable because it increases product cost and component tolerances. High capacitance is a drawback because it narrows the frequency response of the associated node. Therefore, the conventional resistor structures are undesirable for use in gain attenuation.
Because the resistance multiplier circuit of the present invention can provide a large resistance value in a relatively small area, the resistance multiplier circuit can be advantageously applied to an amplifier circuit as a gain attenuator.
Referring to
The first gain point in the amplifier circuit is at the output node of the active load stage (node R2), where the gain is given by the transconductance of the PNP input stage (Q60 and Q61) multiplied by the nodal impedance of the active load output nodes (nodes R1 and R2). In one exemplary fabrication process, the nodal impedance is about 4 Mohms. To attenuate the gain by a factor of 10, a 400 kohms resistance to ground is required at both output nodes R1 and R2 of the active load stage. That is, a 400 kohms resistance is required at each of nodes R1 and R2.
Because a 400 kohms resistance is quite large, it is not feasible to use conventional methods to implement such large resistance. In amplifier circuit 50, the necessary gain attenuation resistances are implemented using the resistance multiplier circuit of the present invention. Specifically, a resistance multiplier circuit 52 is coupled to nodes R1 and R2 to provide the necessary resistances at the respective nodes.
By incorporating resistor multiplier circuit 52 in amplifier circuit 50, a gain reduction by a factor of 10 can be realized without requiring alteration to the structure of the amplifier circuit and without requiring large silicon area to implement. As a result of incorporating resistance multiplier circuit 52 in amplifier circuit 50, the gain of the amplifier circuit is reduced by a factor of 10 and compensation of the amplifier circuit can be now carried out more easily. Furthermore, as a benefit in addition to the gain reduction, the DC currents drawn by resistance multiplier circuit 52 are less than that would be drawn by an equivalent conventional resistor. This is because the voltage drop across the resistors in the resistance multiplier circuit is the output node voltages at nodes R1 and R2 diminished by one base-to-emitter voltage. When conventional resistor structures are used, the DC currents are based on the voltage drops across the resistor structures which are the entire output node voltages at nodes R1 and R2. Because the DC currents drawn by the resistance multiplier circuit are less, offset currents due to the resistor DC currents are also reduced as compared to the case when conventional resistor structures are used.
The above detailed descriptions are provided to illustrate specific embodiments of the present invention and are not intended to be limiting. Numerous modifications and variations within the scope of the present invention are possible. The present invention is defined by the appended claims.
Number | Name | Date | Kind |
---|---|---|---|
5471173 | Moore et al. | Nov 1995 | A |
5734295 | Kagawa | Mar 1998 | A |
20040251965 | Ueda et al. | Dec 2004 | A1 |
Number | Date | Country | |
---|---|---|---|
20040201280 A1 | Oct 2004 | US |